Get measure_id tests to work

[johnzl-777]

This was pretty straightforward but the scf tests I had to rework a bit.

There was a test where I use squin.broadcast.measure in one branch of an scf.IfElse but the problem is that the passes/combination of passes I know to get qalloc to cough up an IList with concrete length inevitably eliminate the scf.IfElse altogether.

My workaround is to just use the 1Q measure which, funny enough, works but only because it immediately creates an IList with a concrete length the measure_id analysis can consume 😅

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[github-actions]

☂️ Python Coverage

current status: ✅

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
9508	8398	88%	0%	🟢

New Files

No new covered files...

Modified Files

No covered modified files...

updated for commit: c781216 by action🐍

[david-pl]

There was a test where I use squin.broadcast.measure in one branch of an scf.IfElse but the problem is that the passes/combination of passes I know to get qalloc to cough up an IList with concrete length inevitably eliminate the scf.IfElse altogether.

@johnzl-777 that shouldn't happen, since broadcast.measure(q) and measure(q[0]) have different side effects: the first one is changing the state for all qubits, where the latter only affects the first one. So, if that IfElse gets eliminated it might be a bug.
I saw that you changed the number of qubits in the test though. So maybe this elimination only happened when using a single qubit, in which case the statements are equivalent?

Also, there was a test failing on Python 3.10, because somehow CSE didn't eliminate a line which threw off the line number of the results. I "fixed" that, by getting the results based on the variable names rather than the line number.

[johnzl-777]

Also, there was a test failing on Python 3.10, because somehow CSE didn't eliminate a line which threw off the line number of the results. I "fixed" that, by getting the results based on the variable names rather than the line number.

Thank you so much for the fix! I guess I should have let CI run its course before going to bed last night ):

As for the broadcast vs 1q measure I think the elimination problem was from the fact I was using a concrete boolean. The way around this is to just have a variable as a condition. And now that I've revisited this I realize that I CAN use Flatten now and get concrete info (: so I'll revise the test 👍